AI technologies are manifesting in many different ways, solving many different problems and creating many new opportunities in payments and financial services.
The only wAI is up!
There are operational applications, that serve middle to back office efficiency goals, and front office developments that create novel solutions and services for the end user.
AI itself being an umbrella term for a new approach and new capabilities of wielding data, will start to yield all manner of sub-categories, as the industry gets to grips with and differentiate between automation and machine learning, and more sophisticated and dynamic neural networks and, indeed, artificial intelligence.
Natural Language Processing (NLP) has been a hugely disruptive area, the advancements of which have been most recently epitomised by ChatGTP.
Use of these technologies represents a paradigm shift for society, not just payments and financial services, governments and authorities having to get up to speed in order to regulate its use but also to improve their own procedures and operations.
To this extent there has been a need for public investment in AI, to help drive forward the right approaches and applications, and to foster talent.
Yet private investment continues to outgrow this due to the huge business benefits to be derived.
Tamsin Crossland, Principal AI Architect, Icon Solutions, shared insights from the field as to the understanding of the benefits AI can bring banks and payments providers, and where they are placing their efforts and budgets in implementing it.
Is the priority for AI more in back of house or front of house for banks, would you say?
To date, there has been a lot of focus on developing AI solutions that increase back office productivity.
For example, improving the customer onboarding experience by using AI to reduce the time taken to process documentation, or to detect fraudulent transactions that may otherwise not have been detected using legacy rule-based systems.
AI can also help to improve banks’ effectiveness when it comes to enforcing sanctions.
In terms of front of house applications, a common use case for AI is chatbots, which enable customers to interact with their accounts using a natural language interface and get quicker answers to queries like “what bills do I need to pay before the next pay day?”.
In some cases, banks have reported that AI-driven chat bots are capable of handling nearly half of all customer queries, with the more complex queries fed by the bank to a human agent to handle.
Although some banks have been more cautious about implementing AI in customer-facing applications, customers have reported high satisfaction rates as a result of reduced waiting times.
Is investment and therefore faith in AI technologies increasing/hampered?
From our work with financial institutions, the programmes that are successful in using AI to deliver real business value, are those that start with a clearly defined business requirement.
For example, to reduce the time taken to onboard customers or to detect more fraudulent transactions.
To produce successful results, AI systems are dependent upon the quality of the organisation’s data being complete, accurate and unbiased.
Where an organisation has a strategy that results in programmes that are driven by business requirements, and the criticality of having high-quality data is understood, AI can and is delivering results that can only increase faith in AI technologies.
Can banks become their best selves without AI?
While banks have managed to survive without AI for most of their history, those that do not take advantage of the opportunity that AI solutions can offer when it comes to improving their services, may struggle to maintain a competitive edge.
AI offers the potential to address so many aspects of banking, from answering customer queries more quickly, to onboarding customers in less time and detecting fraud more efficiently.
The use cases are myriad.
Banks that implement an AI strategy that reflects their business strategy, considers the importance of maintaining complete, accurate and unbiased data whilst also considering the ethical aspects of implementation, are much more likely to become their best selves.
Synechron to shine a light
Next, Anna Milne spoke with Ryan Cox, Head of AI, Synechron to shine a light on the priorities and progress of North American banks and payments providers in rolling out AI technologies, and what effect it is having.
Is the investment paying off or is it too soon to tell?
Equally, when it comes to the UK and private Vs. public investment disparity, where does that leave market competition on a global scale?
What challenges are banks are trying to solve through the deployment of AI?
Banks are strategically deploying AI to address multiple critical challenges in today’s competitive financial landscape.
An important focus is bridging the agility gap with fintech competitors while maintaining robust security and compliance standards, Banks face significant hurdles in unlocking value from large amounts of siloed data across organisations.
AI helps in cleaning, transforming, and analysing this information to drive actionable insights.
They focus on achieving a “zero-ops” mindset through intelligent automation that surpasses traditional RPA limitations.
Unlike rigid rules-based systems, AI solutions provide flexible and adaptive automation that can identify patterns, aggregate trends, and proactively suggest solutions.
Additionally, banks are leveraging AI to modernise legacy systems without investing in increasingly scarce specialised development skills, all while working to meet customer expectations for personalised services and improved experiences.
Are banks that exhibit higher AI maturity performing better?
We are seeing banks with a higher AI maturity outperforming their peers financially and operationally.
The Evident AI Index, which tracks 50 of the world’s largest banks, shows clear leaders like JPMorgan Chase, Capital One, and the Royal Bank of Canada as 2024 leaders in AI maturity.
The three leaders had excellent EPS beats ranging from 9.0% to 19.9% in their most recent earnings reports, showing a correlation between AI maturity and performance.
JP Morgan expects up to $2bn in additional AI-driven revenue from AI initiatives this year, notably from AI-driven fraud prevention.
The bank reports a 15-20% reduction in account validation rejection rates due to AI-powered payment screenings.
McKinsey’s analysis further supports this trend, estimating generative AI could add $200-300bn annually to the global banking sector, representing 9-15% of operating profits.
We’re also seeing banks that are effectively leveraging AI across their knowledge workers and engineering teams are seeing 25-35% productivity improvements, alongside enhanced quality metrics.
How are top-performing banks investing in AI?
Leading banks are making strategic AI investments to transform their operational capabilities and maintain competitiveness.
To achieve success, banks focus on talent acquisition by hiring top AI experts, forming partnerships with tech companies and AI start-ups, investing in research centres, and modernising infrastructure.
These investments aim to increase productivity, reduce costs, and provide personalised services. JP Morgan Chase is leading with substantial AI investments and a comprehensive approach, employing 2,000 AI and machine learning experts and rolling out AI assistants to 140,000 employees to automate tasks like drafting emails and reports.
They are focusing on adopting a multi-cloud approach to modernise infrastructure and harness generative AI responsibly.
Similarly, Morgan Stanley has developed an AI-powered knowledge retrieval assistant for automating meeting notes and action items, allowing advisors to focus on clients.
Creating scalable value with generative AI in highly regulated companies
Successfully scaling generative AI in regulated environments requires balancing innovation and compliance.
Leading financial institutions are achieving this through robust governance frameworks that ensure transparency and accountability in AI-driven decisions.
Humans are kept in the loop to review AI outputs and monitor performance, and audit trails are maintained for inputs and outputs of the model.
Model validation frameworks are critical for evaluating the AI to ensure it continues to perform well if data changes over time or when models are upgraded.
Additionally, building an AI-literate workforce has proven crucial for regulated companies to effectively manage and innovate with these advanced technologies while ensuring compliance with regulatory standards.
With the UK public sector pulling back on AI spending, how is the private sector picking up the slack?
With the UK public sector pulling back on AI spending, it’s down to the private sector to help continue the UK’s prominent position as an AI leader across the globe.
We see significant investment across some of the biggest players in the private sector already.
Amazon Web Services (AWS) has launched a $50m initiative to promote innovation in generative AI within the public sector and has recently invested $35bn in data centres.
JP Morgan spends $15B/year on its 50k technologists, of which AI is becoming increasingly important & they are publicly showcasing their focus on AI.
We’re also seeing a significant increase in spending from major banks and insurance companies, as well as major cloud providers.
It’s important to note that there is healthy collaboration between the private sector, tech firms, universities, and research centres.
The private sector is driving many of these collaborative efforts already and will probably double down now that the UK public sector is pulling back on AI spending.
Private sector AI adaption could actually boost UK GDP by up to 16% in the coming decades.
Why funding is needed in government tech, and what governmental initiatives can accelerate AI investment and applications?
Funding is needed to establish regulatory sandboxes, allowing companies to test AI technology in a controlled environment.
Funding is also needed to address the lag in AI development compared to countries like China and the US which could have national security implications.
Some key initiatives that the government could put in place are:
- Integrate private-sector expertise into public initiatives – e.g. schemes like the UK’s National Cyber Security Centre’s Industry 100
- Workforce investment and education plans – e.g. Initiatives such as the AI Training Act and the CHIPS and Science Act in the US aim to bolster the federal AI workforce by providing scholarships, training programs, and educational resources.
- Infrastructure provider, helping regional innovation hubs to offer low-cost computing infrastructure
- Develop work visas for foreign AI talent
- Democratise access to AI research tools and high-quality data sets
- Leverage AI for smart city initiatives, e.g. Dubai’s AI-powered data centre economy and Estonia’s e-residency program. They demonstrate the potential of AI in improving public services. These projects enhance efficiency, attract investment, and drive economic growth
Can the use of more efficient AI models help moderate the computing power arms race?
Finding enough computing power is a big challenge in AI. Edge computing could be the solution needed in this computing power arms race.
Edge computing means moving more of the computation closer to the user.
This means devices like your smartphone and car can do more heavy lifting locally rather than calling a server potentially sitting hundreds of miles away.
The decentralised compute approach reduces the reliance on massive data centres, which are often energy-intensive.
For instance, Google and Microsoft now consume more electricity than 100+ countries.
Moving compute power to the edge means optimising resource allocation and reducing energy consumption.
An added bonus of processing data closer to its source means a reduced need for costly data transmission to centralised cloud systems.
This lowers latency and bandwidth and allows real-time applications to operate more reliably.
Finally, edge devices are optimised for specific tasks and can utilise smaller language models (SLMs) that require 60x less power to train and run.
The post Q&AI: The only wAI is up! appeared first on Payments Cards & Mobile.